Introduction

The U.S. bike-sharing industry has experienced significant expansion, revolutionizing urban mobility while addressing traffic congestion and environmental sustainability. In major cities, accessible bike-sharing networks are available, offering a variety of bike types including traditional, electric, and smart bikes. Despite a few companies dominating the market, there is room for unique differentiation.

A prominent example is Citi Bike NYC, operated by Lyft, which has established itself as a key player in providing convenient, eco-friendly transportation in New York since its inception in May 2013. Following its success, Citi Bike expanded into Jersey City and Hoboken in 2015. Our team is focused not only on increasing the overall usage of bike-sharing in Jersey City but also on specifically promoting the use of electric bikes. By emphasizing e-bikes, we aim to enhance the sustainability and reputation of Citi Bike NYC, furthering its growth and positive impact on urban transportation.

Research question and approach

To enhance our understanding of the factors influencing the usage rate of Citi Bike, our initial step involved identifying the key elements contributing to the fluctuations in bike usage. A significant pivot in Citi Bike’s business model was the introduction of electric bikes (E-bikes), which marked a notable change in the service’s offerings. Recognizing the potential of E-bikes to attract new users and increase overall bike usage, we tailored our research questions to focus specifically on this aspect.

Our primary research question is: What is the optimal balance of electric bikes and traditional bikes at Citi Bike stations across various regions? We structured our analytical framework to compare E-bikes with traditional bikes in multiple dimensions. This comparison aimed to shed light on user preferences, the impact of E-bikes on ride duration, their role in expanding the user base to different demographics, and the overall effect on the Citi Bike system’s efficiency and appeal. By concentrating on these comparative analyses, we aimed to uncover insights that could guide strategies to boost Citi Bike’s usage rate effectively.

The first question we answered was: what is the difference in top popular routes for electric bikes versus traditional ones? We define a “route” as a distinct pair of start and end stations that define a single trip. This approach will allow us to discern whether there are particular patterns or preferences in the use of E-bikes versus traditional bikes based on the routes users choose, thereby informing us on how to optimally adjust the allocation of each bike type along the routes.

To add depth to our analysis, we also plan to examine the distribution of these popular routes across different days of the week. This aspect of the research will provide insights into how the usage of E-bikes and traditional bikes varies depending on the day. For instance, are certain routes more popular with E-bike riders during weekdays as opposed to weekends? Does the type of bike chosen correlate with the purpose of the trip, such as commuting during weekdays or leisure activities during weekends?

Our second research question delved into the differences in the most popular stations for electric bikes (E-bikes) versus traditional ones. Building on our analysis of popular routes, this inquiry extended to examining how specific stations are utilized differently by E-bike and traditional bike riders. In assessing the popularity of a station, we employed a multifaceted approach, considering several key metrics:

  1. Total Inflow of Bikes: This metric involves counting the total number of bikes, both E-bikes and traditional, arriving at a station. This helps in understanding which stations serve as major destinations.

  2. Total Outflow of Bikes: Contrasting with inflow, this metric measures the total number of bikes leaving a station, providing insights into which stations are common starting points for trips.

  3. Net Change of Bikes Coming Into and Out of a Station: By calculating the difference between the inflow and outflow, we can identify stations that are net ‘givers’ or ‘receivers’ of bikes. This helps in understanding the dynamic balance of bike availability at each station.

  4. Cumulative Number of Bikes: This involves tracking the total number of bikes (both types) passing through a station over a given period, offering a broader view of station usage intensity.

Results: Analysis & Recommendations

For this project we used trip data for the last 2 years, from October 2021 to October 2023.

From the end of 2021 to the end of 2022 electric bike usage followed a similar usage pattern as classic bikes in a lower magnitude. However, since April 2023 the use of e-bikes has decreased, following an opposite trend compared to the growth of classic bike usage.

On average, this is the distribution of ride type:

Riders can pay for a subscription for unlimited bike rides, or they can pay for individual rides.

Members show a preference for classic bikes, with 70% of all classic bike usage attributed to them, while non-members are more inclined towards electric bikes, with members accounting for approximately 60% of all e-bike usage.

Type of bike and usage by different time frames

While there is not a substantial difference in daily bike usage, a higher number of bikes are utilized on weekdays compared to weekends. Notably, Sunday sees the lowest bike usage. Overall, there is no significant fluctuation in the daily usage pattern.

Bike usage reaches its peak between noon and evening (12 PM to 8 PM), comprising almost 40% of overall activity. On the contrary, usage of both classic and electric bikes is minimal during the night, specifically from midnight (12 AM) to 6 AM, representing less than 5% of total.

The usage patterns for both e-bikes and classic bikes are similar. In the morning rush hour (7 AM to 8 AM), they account for 10% of daily bike usage, while the evening rush hour (5 PM to 6 PM) sees a higher usage, contributing to 20% of the day’s total. Therefore, these two hours of peak rush hour collectively represent 30% of the entire day’s bike usage.

Type of bike and trips

On average, rides in electric bikes are longer, both in terms of distance and duration.

The density graph depicting the distribution of ride distances (capped at 5000 meters) for various bike types offers a clear perspective on the average journey lengths associated with each type of bike. It’s noticeable that e-bikes typically have slightly longer riding distances compared to classic bikes, with the most common distance being around 1000 meters. Moreover, the majority of rides, regardless of bike type, do not exceed 2 kilometers. About 6% of rides start and end in the same station, meaning the calculated distance is zero, which explains a somewhat elevated density in 0. It’s important to know that these rides are made exclusively with regular bikes.

The density graph shows that e-bikes are typically ridden at faster speeds than classic bikes. Most e-bike riders travel at speeds ranging from 8 to 15 km/h, while the speed of classic bike riders mostly falls between 5 to 12 km/h. It is also observed that very few bikes, whether electric or classic, surpass the speed of 20 km/h.

Recommendations based on dynamic application

Based on our investigation results regarding the varying patterns of e-bike and classic bike usage from different perspectives, we can formulate recommendations for Citi Bike on optimizing the distribution of these bike types.

Tab 1: Routes

Within the first tab of our dynamic application, named ‘Routes,’ we have delineated the most popular routes taken by Citi Bike rides in Jersey City on different weekdays, visually represented on a map. Users have the option to select the number of most popular routes using a slider and can choose the respective weekdays from the left side. The resulting paths displayed on the map identifies high-demand stations. By analyzing routes with high user traffic, Citi Bike could pinpoint stations along these routes that consistently experience increased demand for bikes. This approach provides our clients with insights into stations where they should allocate a higher number of bikes along these popular routes during different weekdays, considering their popularity. This strategic allocation aims to reduce instances of bike shortages and increase the efficiency of bike allocation in Jersey City.

The dynamic application outputs reveal distinct commuting patterns of popular ride routes across various weekdays. From Monday to Thursday, the most frequented route starts at Marin Light Rail and ends at Grove St PATH, situated in the lower east part of Jersey City, with the highest ride count recorded on Tuesday at 558 rides from Jul 2023 to Oct 2023. The result suggests a commuter-heavy route, likely utilized for work commutes. Conversely, from Friday to Sunday, the most popular route commences at Hoboken Terminal - Hudson St & Hudson Pl and ends at Hoboken Ave at Monmouth St, located in the upper east part of Jersey City, indicating potential leisure or weekend specific travels. The peak usage is observed on Saturday, reaching 379 rides from Jul 2023 to Oct 2023.

The disparity between the routes during weekdays and weekends implies a shift in rider behavior, with weekdays possibly focused on work commutes and weekends geared towards leisurely activities or different destinations. For routes starting from transportation hubs during weekdays (Marin Light Rail, Grove St PATH), it indicate commuter-heavy areas, while routes starting from different locations during weekends(Hoboken Terminal) suggest travel to recreational destinations.

Most Frequented Route (Weekdays) Most Frequented Route (Weekend)

Based on the specific usage patterns observed, we recommend Citi Bike strategically allocate additional bikes during peak times along commuter-heavy routes, notably from Marin Light Rail to Grove St PATH on weekdays, ensuring ample supply, particularly on Tuesdays. Considering the relatively short route distance, Citi Bike could consider allocating more classic bikes at these stations. For weekends, focusing on increased bike availability around Hoboken Terminal would accommodate riders exploring recreational areas. Given the longer route distance, Citi Bike could consider allocating more e-bikes at stations along this route. Moreover, implementing targeted promotions in these areas to enhance social awareness, along with enhancing bike infrastructure along these popular paths, would significantly enhance user experiences.

Tab 2: Stations

In the second tab of our dynamic application named ‘Stations,’ a graphical representation showcases the locations of all stations in Jersey City. This tab shows the number of bikes entering and exiting each station daily, the net change per day, and cumulatively tracks bike movements over the selected time period. You choose to view ebike or classic bike on any given date. This tab provides a visualization of the exact locations of stations on a map, enabling our client, Citi Bike, to monitor the daily traffic at each station.

Notably, the net change of a bike station is calculated using four metrics, which involve subtracting the number of bikes that leave a station (bike out) from the number that arrive (bike in). A negative net change indicates a shortage of bikes at that station, while a positive number suggests a surplus. 

Most stations typically experience single-digit bike inflow and outflow, but stations near PATH hubs like Hoboken and Grove Street often see two-digit figures in bike traffic. These areas are evidently busier than others, indicating a higher demand for bike services. Notably, the Hoboken station is the busiest, primarily due to its role as a major PATH station for New Jersey residents commuting to New York. In the vicinity of Hoboken, there are four closely situated bike stations. Despite their proximity - just a block apart - some stations experience significantly different rates of bike inflow and outflow.

Considering the information provided, we recommend that Citi Bike utilize this tab to gain a clearer understanding of bike usage patterns. This insight will assist in optimizing the distribution of bikes throughout Jersey City, with a particular focus on stations experiencing higher levels of bike inflow and outflow.

Tab 3: E-bikes / Classic bikes

In the third tab of the dynamic application, named ‘E-bikes / Classic bikes,’ we illustrate the change in the number of bikes leaving the station on selected weekdays. Users can select bike types and weekdays on the upper left side, followed by choosing specific station markers displayed on the map. The resulting outcome is presented as a plotted graph, displaying the average numbers of bikes leaving the station during different hours of the day.

Our visualization of the average number of e-bikes leaving the station offers insights into peak hours of bike usage, aiding in strategic resource allocation. Simultaneously, identifying stations experiencing higher bike outflow during specific hours provides opportunities to increase bike stock and promote the usage of e-bikes at those stations to avoid shortages and raise customer awareness.

The graph below illustrates the average number of e-bikes departing from the Marin Light Rail station, previously identified as the most frequented starting point on workdays, particularly on Tuesdays. It reveals peak hours for e-bike outflow at this station around 8 am and 6 pm, aligning with rider behavior analyzed in the ‘Routes’ section. The highest outflow, approximately 8 bikes, occurs at 8 am, followed by a significant drop to an outflow of 2.5 bikes around 9 am, suggesting a potential shortage in e-bikes post-peak commuting hours.



Based on these findings, we recommend that Citi Bike increase e-bike inventory at popular stations during peak commuting hours (8 am and 6 pm) on workdays to meet heightened demand and prevent shortages afterward. Leveraging the functionality in the third tab enables Citi Bike to identify hotspot stations. Considering this, Citi Bike could contemplate expanding or upgrading these stations to ensure sufficient e-bike availability, battery charging facilities, and parking spaces during peak hours. Additionally, conducting rider awareness programs at these hotspots to educate riders about e-bike availability and their advantages would further enhance user experiences.

Tab 4: Grouping Variables

In our ‘Grouping Variables’ tab, the count of e-bike and classic bike rides is depicted in the resulting plot. Users can select the grouping variables from the options provided on the upper left side. The ride counts can be grouped by a single date, a single month, every weekday, or every weekend. The count of classic bikes is represented by the red line, while the count of e-bikes is indicated by the blue line. 

This tab provides Citi Bike with essential insights into rider behavior and usage patterns across different time measures. Through the visualization, Citi Bike gains a comprehensive understanding of daily, monthly, and weekly trends, which is essential for resource optimization by identifying usage patterns.

Tab 5: Density Plot

Our latest density tab offers three variable selections, allowing our customers to visualize the density plot of riding duration in seconds, riding distance in meters, and riding speed in km/h. In this representation, classic bikes are depicted in red, while e-bikes are shown in blue.

As observed in our dynamic application, e-bikes tend to exhibit higher speeds and cover longer distances and durations. This tab provides a visual representation for users to compare how customers ride these two different types of bikes in the city.

Quick-start guide

This section aims to ensure reproducible generation of results.

Data Sources

The trip data utilized for this project was from October 2021 to October 2023 and was retrieved from the following website: https://s3.amazonaws.com/tripdata/index.html

The files corresponding to Jersey City data follow this name pattern: JC_YYYYMM_citibike_tripdata.csv.zip After downloading the files, they must be extracted from the zip format.

Folder Structure

For the dynamic application to run seamlessly, data files and scripts must be organized in the following folder structure:

  • Scripts
    • Dynamic Reporting Engine - Final.Rmd
    • Install_Libraries.R
    • electric_vs_classic.rds
    • shiny_preprocessed_rides.rds
    • station_lat_lng.rds
    • stations_tab.rds
  • Data
    • JC_YYYYMM_citibike_tripdata.csv
    • JC_202207_citibike.tripdata.csv has a typo from the website. After downloading one must manually replace citbike for citibike.

Data Processing

To ensure the dynamic application is user-friendly and requires minimum technical requirements, the data processing is encapsulated within the same script. Because of this, the dynamic application may take a few seconds to run, depending on the computer.

Install Requirements

First time users may need to download software prior to running the dynamic application.

  1. Install R and RStudio: Latest versions of both can be installed from this link: https://posit.co/download/rstudio-desktop/. This guide https://rstudio-education.github.io/hopr/starting.html provides more detailed instructions.
  2. Install required packages: After downloading R and RStudio, open the latter and run the script Install_Libraries.R provided along the deliverables.